artificial intelligence

AI will account for 19% of data center power demand by 2028 – Goldman Sachs

This is AI generated summarization, which may have errors. For context, always refer to the full article.

AI will account for 19% of data center power demand by 2028 – Goldman Sachs
Concerns surrounding the power consumption of artificial intelligence use continue to mount

Ten times the power – that’s how much electricity ChatGPT drains to process a query similarly to a Google Search, according to a Goldman Sachs report.

One ChatGPT request consumes 2.9 watt-hours of electricity, while a Google search only uses 0.3 watt-hours, according to an International Energy Agency report.

Goldman Sachs Research predicts data center power consumption will swell to as much as 200 terawatt-hours annually between 2023 and 2030 just from artificial intelligence technology alone. This increase in data center power demand is estimated to reach 160% by the decade’s end. Furthermore, by 2028, 19% of the power demand for data centers will be coming from AI. 

Similarly, a May 2024 study by researchers from AI startup Hugging Face and Carnegie Mellon University found that, on average, out of 88 AI models, it required 2,907 watt-hours to generate an image from 1000 queries. Text generation used two watt-hours per 1000 inferences, although it can vary depending on the length of the text. 

Even training AI models is a power-hungry endeavor. Kate Saenko, associate professor of Computer Science at Boston University, wrote that, in 2019, the creation of AI models with 110 million parameters was reported to consume as much energy as a trans-American flight for one person. On the other hand, the energy consumption of making GPT-3 was “equivalent” to “123 gasoline-powered passenger vehicles driven for one year.”

The high power consumption from AI use is a scary prospect for sustainability efforts, especially as the technology rapidly grows.

Roberto Verdecchia, an assistant professor at the University of Florence, wrote in a paper on developing green AI solutions that, “In the race to produce faster and more accurate AI models, environmental sustainability is often regarded as a second-class citizen.”

At the World Economic Forum (WEF), OpenAI CEO Sam Altman warned others of AI’s increasing energy consumption needs. As Reuters reported during the WEF, Altman claimed the technology would need an energy breakthrough, proposing nuclear fusion or solar energy as potential solutions. – Rav Ayag/

Rav Ayag is a Tech and Features intern at Rappler. He is an incoming senior at the Ateneo de Manila University in the Bachelor of Fine Arts Creative Writing program. 

This story was vetted by a reporter and an editor.

Add a comment

Sort by

There are no comments yet. Add your comment to start the conversation.

Summarize this article with AI

How does this make you feel?

Download the Rappler App!